The discovery of solute precursors of crystalline materials, such as biominerals, recently challenged the classical nucleation theory (CNT). One emerging method for investigating these early-stage intermediates in solution is dissolution dynamic nuclear polarization (dDNP)-enhanced nuclear magnetic resonance (NMR) spectroscopy. Recent applications of dDNP to calcium carbonate (CaC) and calcium phosphate (CaP) mineralization have demonstrated the feasibility of identifying and tracing very early-stage prenucleation clusters (PNCs). However, the structural details remain difficult to resolve as dDNP is mainly limited to simple onedimensional NMR detection. To overcome this bottleneck, we herein integrate hyperpolarized NMR of PNC with molecular dynamics simulations and quantum mechanical calculations to gain atomistic structural insights into CaP PNCs. By simulating the PNC structures, computing chemical shift parameters, and comparing these to hyperpolarized NMR "fingerprint" spectra, we demonstrate how to derive models of solution-state structural ensembles of PNC, even when very short-lived. With this approach, we find that the Ca/P i ratio inside PNC tends to stay close to 1 independent of pH, while their sizes vary, leading to larger precursors under more basic conditions. At the same time, phosphate speciation within PNC was found to be independent of pH, as only monohydrogen phosphates participated in PNC formation. This latter feature also entailed a pH-independent local atomistic arrangement of phosphates coordinating a Ca(II) center, leading to constant Ca 2+ -P i distances of ∼3 and ∼3.6 Å. These ion-to-ion distances agree with those found inside solid CaP phases such as brushite, octacalcium phosphate, or hydroxyapatite - a feature hinting toward the templating function of PNCs. Thus, our method (i) extends the methodological scope of hyperpolarized NMR by complementing one-dimensional fingerprint spectra with full structural models and (ii) sheds light on key intermediates that have been experimentally underexplored.